STOCKS | Who are the people creating the PROFIT EXPECTATIONS that drive stock prices up--and down?

IT'S EARNINGS-reporting season and the talking heads on CNBC's "Squawk Box" are dutifully relaying the numbers to viewers. PairGain Technologies reports 2 cents per share--"in line," says laid-back Joe Kernen. Genial host Mark Haines tells us that Viant "beat the Street." Bubbly Liz Claman tells of CEO Jill Barad's resignation even as Mattel "misses estimates yet again."

If you're new to the world of stocks, you are forgiven for rolling your eyes at this babble. And yet the subject of the patter--actual reported profits compared with what the analysts who follow the companies forecast--is perhaps the single-most-important factor driving share-price performance over the short term, and plays an important role in longer-term performance as well. This should come as little surprise to anyone: who's owned a stock that "missed the consensus" by a few pennies per share and, as a result, tumbled 30%, 40% or more in a day.

For better or worse, investors obsess over consensus profit estimates. Now more than ever, it's critical that you know what professionals expect from the companies you hold. "Investing is based on expectations for the future," says Edward Keon, chief quantitative analyst at Prudential Securities. If expectations aren't met or if they change, a company's share price usually changes precipitously.

Looking into the future has long been an integral part of picking stocks. But the inordinate attention paid to consensus earnings expectations is relatively new. Ten years ago, Peter Lynch didn't even mention the concept when he wrote One Up on Wall Street, his stock-picking primer.

And yet one firm has compiled and distributed consensus estimates since the early 1970s. I/B/E/S International (the abbreviation stands for Institutional Brokerage Estimate System) remains in business today and has been joined by such rivals as First Call/Thomson Financial and Zacks Investment Research. In the past decade investors began to follow these numbers avidly because they noticed that stock prices of companies that fail to meet consensus earnings estimates tend to drop, while these that "surprise on the upside," in WaLl Street geek-speak, go up. "So by the early part of the 1990s, the consensus numbers grew from being a nice-to-have reference tool to being an essential part of the investment process," says Keon, a former research director at I/B/E/S.

Collecting the data

THE THREE major data gatherers--First Call, I/B/E/S and Zacks--emphasize somewhat different markets and collect estimates from slightly different sources. But all basically opera-e in the same fashion. In simple terms, their work falls into two areas. First, they collect earnings estimates from analysts at brokerage firms and independent research houses. (Analysts working for money-management firms are not polled.) The analysts typically supply earnings estimates for the next four quarters, estimates for the current year and the next fiscal year, and the expected rate of earnings growth over the next three to five years.

All three companies strive for consistency and timeliness. For instance, they all try to calculate averages for operating earnings--profits before such one-time events as restructuring charges and asset sales. "We spend a lot of manpower making sure that all the estimates are on the same basis," says Charles "Chuck" Hill, director of research at First Call. Where there is controversy over a particular issue, "we go with what the majority of the analyst community wants to do," says Hill.

In calculating the average, an estimate from analyst at Goldman Sachs carries no more weight than that of an analyst at a small brokerage in Podunk, U.S.A. If, however, an estimate seems out of kilter with other estimates, it may be excluded from the consensus. Also, says Rick Pucci, chief operating officer of I/B/E/S, an estimate may be excluded if an analyst is slow to react to a company's statement regarding future profits. At Zacks, says Spiro Papas, manager of the brokerage-research data group, an estimate is ignored if the analyst neither updates nor reiterates it within 120 days of first issuing the number.

Estimating a company's profits is part science, part art. Hill, for one, thinks the consensus numbers provide a reasonably accurate picture of the outlook for corporate America. Looking at companies in Standard & Poor's 500-stock index over the past five years, 19% of profit reports have been "on target to the penny," reports Hill, and another 40% have come within 5% of the consensus estimate.

But these numbers need to be taken with the proverbial grain of salt. In the early 1990s, with the economy just coming out of recession, consensus earnings were consistently above actual results and corporations were moved to provide better guidance to Wall Street analysts. That sometimes means talking down analysts' forecasts, making it less likely that a company will fail to make its numbers--and more likely that the company will deliver a positive earnings surprise.

Trumpets and whispers

A GROWING number of companies trumpet in advance of their formal earnings report that profits will either trail or exceed forecasts. Some 90% of "preannouncements" disclose negative news. By revealing bad news early, managements can insulate themselves from lawsuits by investors who bought shares between the time company officials first got wind of punk earnings and the time the figures were formally released. After a preannouncement, analysts usually revise their estimates, making them more accurate than they might have been.

Recently, analysts have added new meaning to the expression "the word on the Street": the "whisper number." This is a second, normally higher, estimate that is rumored to represent what an analyst really thinks a company will earn in a given quarter. An analyst will relay his whisper number to his firm's larger clients--generally mutual and pension funds. William Keithler, manager of Invesco Technology fund, frequently receives such calls from Wall Street analysts. "Most of the significant stocks I own have whisper numbers," says Keithler. Individuals can try their luck at getting whisper numbers for some companies online (www.earningswhispers.com).

Whisper numbers are controversial. Companies sometimes meet earnings expectations only to see their stock price fall because profits trailed the whispers. Many observers question the ethics of the practice. Does the whispered have insider information? Is it fair to whisper to some clients but not to all? Moreover, no one is held accountable, at least in public, for whispered forecasts. John Karns, manager of Westcore Small Cap Growth, says that short-sellers, who benefit from falling share prices, may plant artificially high whisper numbers to increase the likelihood that stocks will fall because companies don't meet the inflated estimates. Some experts suggest that whisper numbers are nothing more than a penny or two added to estimates of companies that often deliver positive surprises.

What to make of it

OUR ADVICE: Ignore the whisper estimates and the attendant controversies. Instead, concentrate on these two strategies, which have proved effective over the years:

* Beware of cockroaches. The cockroach theory holds that just as you rarely find one roach in the cupboard, you will rarely find a single earnings surprise, good or bad. Behind the theory is the idea that an earnings surprise can often signal a change in the long-term outlook for future earnings.

Many investors say they're inclined to sell the stocks of companies that surprise by failing to meet estimates. Says Mark Weiner, a manager with the One Group of funds: "If your stock disappoints--even if it's down 40% on the first trade--four times out of five the first trade is the best trade for the seller. You're better off selling on the bad news."

* Watch for estimate revisions. Sometimes the trend is more important than the current number. Research has shown that the stocks of companies whose estimates are revised upward usually perform well, while stocks of companies with downward revisions tend to be mediocre performers. "If you can track consensus estimates over time and see those numbers rising or falling, that's a tradable strategy," says Michael Caplan of David L. Babson & Co., a money-management firm in Cambridge, Mass.

Where to find the numbers

LISTS OF recommended companies whose estimates are being boosted by analysts are available from a variety of sources. Brokerage firms that employ quantitative analysts, such as Merrill Lynch, Prudential Securities and Robinson-Humphrey, in Atlanta, produce such lists for clients.

If you want to be your own quant, you can get plenty of consensus-earnings data on the Web. Here's what the three major providers offer.